← Back to results
ENहिं

ElevenLabs Scribe v2

Speech-to-text model from ElevenLabs designed for multilingual transcription workflows.

verifiedRecently verifiedSource: elevenlabs-model-docs
Health low (28)

Best for

  • • Call-center logs, podcasts, and multilingual transcription pipelines.

Limitations

  • • Domain-specific jargon requires custom vocabulary adaptation.
  • • Noisy multi-speaker audio may need pre-cleaning for best accuracy.

Use carefully when

  • • Legal-grade transcripts without human review.

Quickstart

  1. Use chunked uploads and speaker labeling for meeting-note pipelines.

Setup checklist

  • • API key required: Yes
  • • SDK quality: high
  • • Self-host difficulty: easy

Health Meter

  • • Setup complexity: 35
  • • Safety & misuse risk: 28
  • • License/compliance risk: 18

• Good baseline for controlled deployment.

Capabilities

  • speechToTexttrue
  • supportedLanguages99
  • diarizationtrue
  • timestampedOutputtrue

Benchmarks

languageCoverage
99
transcriptCompleteness
87.4
diarizationStability
83.2

Community reviews

1 reviews • avg 4

Good multilingual transcript quality

★ 4

Handles mixed Hindi-English calls well after light audio cleanup.

Samples

Transcript excerpt

Hindi-English meeting transcript with timestamps.

Last verified: 25/2/2026, 3:45:28 am • Source: https://elevenlabs.io/docs/models/
AI Bazaar